Multi-agentic AI for Automatic Course Syllabus Generation using LangGraph

Preparing course syllabus is a time-intensive process that requires careful alignment with educational goals, learner backgrounds, and current technological trends. This paper presents a multi-agent AI framework for automatic course syllabus generation, leveraging the LangGraph framework to orchestrate collaborative workflows among large language model (LLM) agents. To address the persistent challenge of hallucination in LLM-based content generation, our system employs an iterative agentic workflow, augmented by expert evaluation from an LLM. We implement GPT-40 as the core LLM, selected for its strong performance and efficiency. The proposed system generates a syllabus through stages including persona analysis, receive human feedback, retrieve standard curriculum, syllabus creation, expert review, and syllabus refinement. Each stage is modeled as modular nodes within LangGraph. We demonstrate the effectiveness of this approach with a case study on a Database Systems course. Evaluation by LLM-simulated domain experts, including database systems educator, data engineering specialist, and data security analyst, highlights the quality and relevance of the generated syllabus. Our findings suggest that multi-agentic systems have the potential to structure courses in a way that aligns with user requirements.